How to decide the parameters of GAMMA and NOISY_GAMMA ?

91 views
Skip to first unread message

Moon Moon

unread,
Jan 7, 2023, 4:35:08 AM1/7/23
to cp2k
Dear CP2K community,
I am doing AIMD using second-generation Car-Parrinello MD.
But I am confusing with the value of GAMMA and NOISY_GAMMA.
Reading related papers, they all said like the following:

微信截图_20230107182222.png
So, I guess these two parameters should be revised in preliminary  run until the system temperature oscillates slightly around the target temperature, is it right ?

And I am also confused with the figure 1 in the paper "Effificient and Accurate Car-Parrinello-like Approach to Born-Oppenheimer Molecular Dynamics".  Why is the energy curve of SGCP so similar with that of BOMD, rather than oscillating arbitrarily near a certain horizontal line ?
QQ浏览器截图20230107183132.png

Any comments will be appreciated, thank you.


Thomas Kühne

unread,
Jan 7, 2023, 5:32:46 AM1/7/23
to 'Dorothea Golze' via cp2k
Dear Moon, 

in the 2nd image, the same trajectory is retraced. I.e. the trajectory is 1st recorded using SGCP and then retraced and the 
energy fully minimized. Also, be aware that this is the potential energy only and not the total energy. 
Wrt. how to adjust the parameters, one way is to scan different values for NOISY_GAMMA (= gamma_D of the paper) and 
then to overlay with a small conventional Langevin GAMMA to correct for remaining accuracies of the former. 

Best, 
Thomas Kühne

Am 07.01.2023 um 10:35 schrieb Moon Moon <yue...@gmail.com>:

Dear CP2K community,
I am doing AIMD using second-generation Car-Parrinello MD.
But I am confusing with the value of GAMMA and NOISY_GAMMA.
Reading related papers, they all said like the following:

<微信截图_20230107182222.png>
So, I guess these two parameters should be revised in preliminary  run until the system temperature oscillates slightly around the target temperature, is it right ?

And I am also confused with the figure 1 in the paper "Effificient and Accurate Car-Parrinello-like Approach to Born-Oppenheimer Molecular Dynamics".  Why is the energy curve of SGCP so similar with that of BOMD, rather than oscillating arbitrarily near a certain horizontal line ?
<QQ?????20230 107183132.png>

Any comments will be appreciated, thank you.



--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/a756c5c4-287e-4e73-b133-c4cc410c91bbn%40googlegroups.com.
<QQ?????20230 107183132.png><微信截图_20230107182222.png>



==============================
Thomas D. Kühne
Dynamics of Condensed Matter
Chair of Theoretical Chemistry
University of Paderborn
Warburger Str. 100
D-33098 Paderborn
Germany

Message has been deleted

Moon Yue

unread,
Jan 12, 2023, 9:57:43 PM1/12/23
to cp2k
Dear  Kühne

Thank you for your kind reply.
Do you mean that corresponding BOMD values should be calculated for energy and force using the configurations produced and recorded using SGCP at different times ?
If doing in that way, Why is there continuous line of BOMD instead of dotted line ?

Best regards,
Yue

Reply all
Reply to author
Forward
0 new messages